One Year Lead Time Experimental Streamflow Forecasts for the Columbia River at The Dalles
2006 water year forecast (ilast updated 10.18.05)
This website presents experimental long-lead (~1-year lead-time) streamflow forecasts for the Columbia River at The Dalles (figure 1) in the U.S. Pacific Northwest. This forecast of monthly streamflow throughout the water year (September-October) is typically made in June preceding the water year and is only updated as necessary during the summer to reflect significant changes in climate conditions. The forecast is not updated during the winter (for seasonally adjusted streamflow forecasts, see our experimental quasi-operational western U.S. streamflow forecasts).
The experimental long-lead forecasts are produced using forecasted El Niño/Southern Oscillation (ENSO) conditions for the upcoming winter to condition resampling of the historical meteorological data used to drive a macro-scale hydrologic model initialized with current hydrologic conditions. In these forecasts, about 1/3 of the climatological streamflow probability distribution is eliminated and the forecast lead-time is increased by about six months compared to existing statistical forecasts based on observations of snowpack (Hamlet and Lettenmaier 1999).
Hamlet et al. 2002 quantified the economic value of long-range streamflow forecasts for Columbia River hydropower. The bottom line: long-range forecasts are worth about $150 million/year. This paper recently won an award for the best practice oriented paper published by the ASCE Journal of Water Resources Planning and Management in 2002.
Climatological Basis for Long-Range Streamflow Forecasts in the Columbia River Basin
The choice of the Columbia River basin as our primary area of study was originally based on the river's importance to the region. In hindsight, the basin proved to be an excellent choice on climatological grounds as well. The Columbia is strongly influenced by both by ENSO and by decadal scale climate variability associated with the Pacific Decadal Oscillation (PDO). This figure shows the effects of PDO and ENSO on naturalized summer (April-September) streamflows at The Dalles, Oregon. Note the consistent shifts in the long term mean associated with warm and cool PDO epochs, and the consistent positioning of the river's response to warm and cool ENSO winters.
The consistent signals over the 20th century associated with these two climate phenomena are the basis for the retrospective resampling streamflow forecasting methods we have developed for the Columbia River basin.
Interpretation of Ensemble Forecasts of Winter ENSO State
Relatively accurate long-range forecasts of the mid-winter ENSO state (warm, neutral, cool) are typically available in June preceeding the forecasted water year (i.e.~six months lead time). Long-term assessments of ENSO forecast skill are problematic due to difficulties in accounting, retrospectively, for the sub-surface ocean data provided by the Tropical Atmosphere Ocean (TAO) buoy system. For the past five years, however, our interpretation of ensemble ENSO3.4 forecasts from National Center for Environmental Prediction (NCEP) and the European Center for Climate Prediction (ECCP) on about June 1 has correctly identified the mid-winter ENSO state for the coming winter in four out of five water years (1998, 1999, 2000, 2002). Water year 2001 was incorrectly forecast as ENSO neutral in June, but proved to be a mild La Niña. Methods for categorizing winter ENSO state are given here.
Applications to Water Resources Management in the Columbia River Basin
A discussion of the potential use of long-range streamflow forecasts for water management and a retrospective assessment of the economic worth of forecasts for the Columbia River hydro system is presented in Hamlet and Lettenmaier (1999) and Hamlet et al. (2002).